A kind of tolerance delay network method for routing based on fuzzy reasoning
Technical field
It is specifically a kind of in transmission success rate and average party in delay face the present invention relates to computer communication technology field
Effect, the tolerance delay network method for routing based on fuzzy reasoning are obtained.
Background technology
Routing algorithm is the primary premise for influencing whole communication system efficient communication.Traditional routing algorithm is assumed in data
Before bag transmits, existing one path end to end between source node and destination node, and such path is not always necessarily
It is available.In this research, node usually has faster speed and larger inertia in three dimensions flight, in addition possibility portion
In hostile environment, uncertain external condition and deliberate human interference all may cause the connection just stablized to become for administration
It is unreliable and be difficult to rely on.During distributed tasks carrying, node individual is also easy to be separated from node cluster.Cause
This, the generation of available communication link is always uncertain, it is necessary to which a kind of adaptive dynamic routing algorithm maintains normal net
Network communicates.
Fig. 1 shows be once simply to dispose process schematic.8 aerial nodes are divided into two groups and hover over different height
Degree, such as Fig. 1(a).The centre of sphere of each ball illustrates the node individual of suspension.The radius of a ball represents the half of individual communications distance.When
When two balls are intersecting, distance meets communication condition between representing individual, so as to generate an available link connection.This cube
After body structure receives " ring-type perception " order from base station, it will perform the process of self planning deployment, such as Fig. 1(b)-
Fig. 1(f).Despite a simple deployment order, it can be seen that air carrier is more prone to shape compared with the mobile unit on ground
Into unstable topological structure.In addition to the internal causes such as inertia that aerial uncontrollable high speed is brought, the error of GPS device
Offset and unexpected air agitation environment are likely to have an impact topological structure.
Delay-disruption tolerant network(Delay/ Disruption Tolerant Network, DTN)Proposition be exactly for understand
Certainly problems:(1)Not always there are available end-to-end path between data source and destination;(2)Between node pair
The maximum round trip time it is sometimes too long;(3)Packet loss rate is larger end to end.Kenvin Fall are carried in SIGCOMM ' 03
This kind is gone out towards " the message-oriented covering layer architecture of tolerance delay " of limited network, abbreviation Delay Tolerant Network.DTN nets
Network is suitable for those since Node distribution is sparse, or topologies change caused by quick movement is violent, and end-to-end path is not
Stable scene.
DTN is to be based on message(Message)Exchange.Its architecture is different from traditional Internet system knots
Structure, introduce one so-called " binding layer ", is solved the problems, such as with this in Delay Tolerant Network.DTN is towards asynchronous message transfer
Nerve of a covering, binding layer operation is in the transport layer of heterogeneous networks, there is provided the function similar with Internet pass, but have very big
Difference, because it is gathered in empty message forwarding rather than packet switch.
Since traditional procotol is all based on communication mechanism end to end, that is to say, that two nodes are starting to communicate it
It is preceding to have erected a complete communication link, while these agreements all assume two-way time of the data frame between node not
The too long and data packet of energy or the Loss Rate of packet are smaller.But in new tolerance delay network, since node is constantly fast
Speed is mobile, and topological structure is in an unsure state, and node is difficult to erect complete communication link when communicating.Therefore data
Transmission can only use " storage-carrying-forwarding "(Store-Carry-and-Forward)Pattern carry out, similar to Rong Chirong
Circuit network(Delay/Disruption Tolerant Network).
Delay Tolerant Network regards the disconnection of link and the segmentation of topological structure as a kind of agreement premise and rule, rather than one
Kind occasional irregularity.The sensor node of aerial mobile needs to communicate under the unstable adverse circumstances of link.Mobile node enters
In the radio-frequency region of other side, the chance of exchange data packets there has been.Whether routing module detection node has received the number in itself
According to bag or the inherently destination of data packet.If so, then selection is abandoned or received;Otherwise, turned according to routing policy
Hair.Fuzzy routing algorithm proposed by the present invention(Fuzzy Routing Algorithm, FRA)It is also based on the back of the body of Delay Tolerant Network
Scape, it is proposed that new routing policy.
The content of the invention
The technical assignment of the present invention is to solve the deficiencies in the prior art, there is provided one kind reduces operation cost of enterprises, saves and deposit
Store up the tolerance delay network method for routing based on fuzzy reasoning in space.
The technical scheme is that realize in the following manner, a kind of tolerance delay network road based on fuzzy reasoning
By method, it realizes that process is:
First, each node Dynamic Maintenance a neighbor node table NNT, stores the neighbour currently in node communication range
Node is occupied, and it is ranked up according to the size of heuristic information value, whenever there is the into or out communication of oneself of node
Scope, the table all can be automatically according to the Policy Updates of fuzzy inference system;
2nd, a data packet is often produced in network, source node just possesses the data packetA backup, hereafter whenever connecing
When triggering is raw, before present node is selected in its NNTA node carries out the forwarding of data packet as receiving node;If saved in table
Point sum, then ownThe chance that a neighbor node will all be forwarded, in whole process, each node transmissionThe backup of a data packet, the number of copies of oneself are updated to;
If the 3rd, node itself number of copiesLess than neighbor node number to be transmitted, then selected from NNTA maximum
Neighbor node, each node transmits the copy of a data packet;WhenWhen, then take direct delivery strategy, i.e., until
Meet the transmission that destination node just carries out data packet.
In the transmittance process of the step 2, if there is new neighbor node to be moved in the communication range of itself, and its energy
The top N of NNT is discharged into, then NNT dynamics update, and node stops to thePosition neighbours, i.e., the heuristic letter in repeating process
The transmission of the node of breath value minimum, and it is transmitted to new node;If receiving the neighbor node of data packet due to change in topology
The reason for have ceased current ongoing transmission, then sending node dynamic renewal NNT, originally in theThe node of position
Move forward toPosition simultaneously obtains repeater-transmitter meeting.
The rule of fuzzy inference system, which is established, in the step 1 comprises the following steps:
1)Input/output variable is blurred:
Two parameters easily obtained are chosen as the metric of localized network state, part is then obtained according to approximate reasoning
Heuristic information, and as the judgment criteria of routing decision;
Output using heuristic information HI as fuzzy inference system, the heuristic information from high to low, are divided into 5
Level:It is very high, high, medium and low, very low;
2)Establish fuzzy control rule:The rule uses Mamdani models, and each control rule is all a fuzzy bar
Part sentence;
3)Fuzzy output sharpening:Median method is chosen to obtain the exact value of heuristic information.
The step 1)In the parameter that easily obtains refer to concentration class CD and relative velocity RVD, wherein concentration class CD is illustrated
Local density centered on both candidate nodes, the value of concentration class CD is total for the neighbor node of the node, the concentration class
Selection process is:The domain of CD is determined first;Then trapezoidal membership function is used, CD is classified, and determines every grade of correspondence opinion
Fuzzy set on domain is that is, basic, normal, high;Relative velocity represents the space angle of speed between node, wherein sending node and candidate
The velocity angle of node is smaller, and RVD values are smaller, and when direction is completely on the contrary, during into 180 degree, RVD values reach maximum, this is opposite
The selection process of speed is:Node moves to directional velocity their angle is calculated in same plane, RVD can be obtained
Value, the RVD values use triangular membership, are 5 grades by RVD points, the fuzzy set on every grade of corresponding domain, i.e., very low, low,
It is middle and high, very high.
The step 2)Middle fuzzy condition statement refers to:
If CD is low, and RVD is very low, then HI is very low;
If CD is low, and RVD is low, then HI is very low;
During if CD is low, and RVD is, then HI is low;
If CD is low, and RVD is high, then during HI is;
If CD is low, and RVD is very high, then HI is high;
If during CD is, and RVD is very low, then HI is very low;
If during CD is, and RVD is low, then HI is low;
During if CD is, and during RVD is, then during HI is;
If during CD is, and RVD is high, then HI is high;
If during CD is, and RVD is very high, then HI is very high;
If CD is high, and RVD is very low, then HI is low;
If CD is high, and RVD is low, then during HI is;
During if CD is high, and RVD is, then HI is high;
If CD is high, and RVD is high, then HI is very high;
If CD is high, and RVD is very high, then HI is very high.
The caused compared with prior art beneficial effect of the present invention is:
A kind of tolerance delay network method for routing based on fuzzy reasoning of the present invention is suitable for the communication of new dynamic environment
Agreement, to shield the network details of bottom, provides unified interface, for the collaboration application on upper strata in order to node preferably
The interaction of message context is carried out, and not cause task distribution failure because of high coupled characteristic or perform deviation occurs, it is practical
Property is strong, easy to spread.
Brief description of the drawings
Attached drawing 1 realizes flow diagram for the present invention's.
Attached drawing 2 is network structure transition diagram of the invention.
Attached drawing 3 calculates schematic diagram for the concentration class of the present invention.
Attached drawing 4 calculates schematic diagram for the relative velocity of the present invention.
Embodiment
It is following detailed to a kind of tolerance delay network method for routing work based on fuzzy reasoning of the present invention below in conjunction with the accompanying drawings
Describe in detail bright.
As shown in Figure 2, a kind of tolerance delay network method for routing based on fuzzy reasoning is now provided, it realizes process
For:
When node mutually into other side less radio-frequency in the range of, there has been transmit forwarding data packet chance.Where is selection
One neighbor node is the core content of routing algorithm as next-hop.In the utility model algorithm, calculated by fuzzy inference system
Obtained heuristic information value is forwarded to the selection of node using as with reference to standard.
Each node Dynamic Maintenance a neighbor node table(Neighbor Nodes Table, NNT), store current
Neighbor node in node communication range, and it is ranked up according to the size of heuristic information value.Whenever have node into
Enter or leave the communication range of oneself, which all can be automatically according to the Policy Updates of fuzzy inference system.
A data packet is often produced in network, source node just possesses the data packetA backup.Hereafter whenever there is contact
During generation, before present node is selected in its NNTA node carries out the forwarding of data packet as receiving node;If table interior joint
Sum, then ownThe chance that a neighbor node will all be forwarded.In whole process, each node transmissionThe backup of a data packet, the number of copies of oneself are updated to。
In transmittance process, if there is new neighbor node to be moved in the communication range of itself, and it can be discharged into the preceding N of NNT
Position, then NNT dynamics update, node stops to thePosition neighbours(The section for the heuristic information value minimum being in repeating process
Point)Transmission, and issue new node;Similarly, if receiving the neighbor node of data packet due to change in topology
Reason have ceased current ongoing transmission, then sending node dynamic renewal NNT, originally in theBefore the node of position
Move toPosition simultaneously obtains repeater-transmitter meeting.
If node itself number of copiesLess than neighbor node number to be transmitted, then selected from NNTThe neighbour of a maximum
Node is occupied, each node transmits the copy of a data packet.WhenWhen, then direct delivery strategy is taken, i.e., until meeting
Destination node just carries out the transmission of data packet.
Wherein, the rule for the fuzzy inference system mentioned in above-mentioned steps, which is established, to be comprised the following steps:
First, the blurring of input/output variable.
In the mobile network under strong interference environment, local heuristic information, to a certain extent than historical experience more
It can reflect provisional link condition, it reflects the availability of each potential path instantly.If neighbor node has bigger
Heuristic information value, being considered as it has the transmission reliability of higher, and vice versa.Fuzzy set theory is introduced to solve the problems, such as,
Select the evaluation metric of heuristic information, it is necessary to meet two conditions:
(1)The metric must be able to the reliable trust degree for reflecting node to be selected.I.e. why requestor believes the candidate solution
There is the transfer probability of higher compared with the candidate of itself or other solution spaces.
(2)The metric must be easy to get and easily calculate.The calculating of node itself and perception are limited, spend big
It is unadvisable in the acquisition of metric to measure resource.
Therefore, metric of two parameters easily obtained as localized network state is have chosen, then according to approximate reasoning
Local heuristic information is obtained, and as the judgment criteria of routing decision.
The two parameters refer to concentration class and relative velocity, wherein:
Concentration class CD is the abbreviation of Crowd Density.Concentration class illustrates the local density centered on both candidate nodes,
For convenience, value is total for the neighbor node of the node.By taking Fig. 3 as an example,It is the source node for carrying forwarding data packet, it is existing
Needed at itWithBoth candidate nodes are selected between two neighbor nodes.Naturally, the candidate of more neighbor nodes is possessed
Node has more chances and message packets is transmitted to destination node, because it has more selections.In figure 3,CD
Be worth for 1 andCD values be 3.Therefore fromFrom the perspective of,Possess more preferable heuristic information.Determine CD's first
Domain.In 2000 × 2000 plane, emulated using random walk and random waypoint mobility models, really
Deckle dividing value is 9.It is 3 grades by CD points using trapezoidal membership function, the fuzzy set on every grade of corresponding domain [0,9]:It is low(Low,
It is abbreviated as L), in(Medium, is abbreviated as M), it is high(High, is abbreviated as H).
Corresponding to membership function is:
,,。
Relative velocity RVD is the abbreviation of Relative Velocity Direction.Relative velocity is defined as between node
The space angle of speed.Sending node and the velocity angle of both candidate nodes are smaller, and RVD values are smaller.When direction completely on the contrary, into
During 180 degree, RVD values reach maximum.Because the directional velocity of node pair is closer, both candidate nodes are met within a period of time in the future
To mobile node also there is bigger possibility to be met in itself by sending node.And ideally, it is intended that by message
Copy is transmitted to neighbor node that those can encounter different nodes to increase transmission success rate.Fig. 4 illustrates such a example
Son.It is the source node for carrying forwarding data packet,WithIt is both candidate nodes.It is apparent thatRVD values ratioBigger,
So as to obtain forwarding qualification.It is in this way, finalWithHave the opportunity to obtain the message copy.
Node moves to directional velocity their angle is calculated in same plane, RVD values can be obtained.Using triangle
Shape membership function, is 5 grades by RVD points, the fuzzy set on every grade of corresponding domain [0,180]:It is very low(Very low, are abbreviated as
VL), it is low(Low, is abbreviated as L), in(Medium, is abbreviated as M), it is high(High, is abbreviated as H), it is very high(Very high, brief note
For VH).Corresponding to membership function is:
,,
,,
。
Outputs of the heuristic information HI as fuzzy inference system, is the abbreviation of Heuristic Information.This is opened
Hairdo information is the important measure factor in routing algorithm.From high to low, 5 grades are divided into:It is very high(Very high, brief note
For VH), it is high(High, is abbreviated as H), in(Medium, is abbreviated as M), it is low(Low, is abbreviated as L), it is very low(Very low, letter
It is denoted as VL).Corresponding to membership function is:
,,
,,
2nd, fuzzy control rule is established.
Control rule is the core of fuzzy system, its form is " if ... ... ".It is defeated in view of two inputs and one
Go out, its rule is generally represented with compound fuzzy condition statement, such as:
If CD is H, and RVD is VH, then HI is VH.
If CD is L, and RVD is M, then HI is L.
The fuzzy inference system employs Mamdani models, and each control rule is all a fuzzy condition statement, institute
Regular composition MULTIPLE COMPOSITE fuzzy condition statement.The control rule table of use is as shown in the table.
If CD is low, and RVD is very low, then HI is very low;
If CD is low, and RVD is low, then HI is very low;
During if CD is low, and RVD is, then HI is low;
If CD is low, and RVD is high, then during HI is;
If CD is low, and RVD is very high, then HI is high;
If during CD is, and RVD is very low, then HI is very low;
If during CD is, and RVD is low, then HI is low;
During if CD is, and during RVD is, then during HI is;
If during CD is, and RVD is high, then HI is high;
If during CD is, and RVD is very high, then HI is very high;
If CD is high, and RVD is very low, then HI is low;
If CD is high, and RVD is low, then during HI is;
During if CD is high, and RVD is, then HI is high;
If CD is high, and RVD is high, then HI is very high;
If CD is high, and RVD is very high, then HI is very high.
3rd, fuzzy output sharpening.
The output obtained by approximate reasoning is the fuzzy set on domain, to turn to and determine that number could be used as heuristic information
Value compares the quality of both candidate nodes.In numerous ambiguity solution methods, median method COA, that is, center of area are chosen to obtain
The exact value of heuristic information.
With the increase of CD and RVD values, node obtains the HI values of bigger;That is, the degree of belief for obtaining higher is made
Forwarding data packet is gone for forward node.
The foregoing is merely the embodiment of the present invention, and within the spirit and principles of the invention, that is made is any
Modification, equivalent substitution, improvement etc., should all be included in the protection scope of the present invention.